Principal Investigator: Dr Kari North
University of North Carolina at Chapel Hill, Epidemiology
137 East Franklin Street, Suite 306, Chapel Hill NC 27514, United StatesTags: 25953, TWAS
1a: The proposed research will leverage novel transcriptome-wide association study (TWAS) approaches to impute predicted gene expression in the UK Biobank cohort. We will then assess association of predicted gene expression with a number of quantitative cardio metabolic traits relevant to diseases like diabetes, heart disease, and stroke. This analysis will allow us to compare the results from novel and published TWAS methods in a large, well-phenotyped cohort. Along with this methods development goal, we will hopefully validate known genes related to cardio metabolic traits and may discover new genes which have been missed by single variant genetic association analyses.
1b: This research will help test a new genetic analysis method, which can be applied to understanding the genetic risk factors for a diverse range of disease-related traits. We will also apply these methods to attempt to discover novel genetic risk factors for important cardio metabolic risk factors relevant to common, complex diseases. Genetic risk factor discovery can lead both to improved risk prediction and novel understanding of disease mechanisms, potentially guiding development of new therapies.
1c: There are a number of publically available datasets (for example, the Genotype-Tissue Expression (GTEx) project) which relate genotypes at millions of sites across the genome to measured gene expression in a number of different tissues, such as adipose, brain, or muscle. These datasets can be used to estimate gene expression in large numbers of individuals where gene expression has not been assessed. We will test both a novel and existing method for estimating gene expression. We will then test to see if this predicted gene expression is associated with changes in a number of cardio metabolic traits.
1d: We plan to include all UK biobank participants with available genome-wide association study data and data for any of the requested cardio metabolic phenotypes.
The proposed research will leverage novel transcriptome-wide association study (TWAS) approaches to impute predicted gene expression in the UK Biobank cohort. We will then assess association of predicted gene expression with a number of quantitative cardiometabolic traits relevant to diseases like diabetes, heart disease, and stroke. This analysis will allow us to compare the results from novel and published TWAS methods in a large, well-phenotyped cohort. Along with this methods development goal, we will hopefully validate known genes related to cardiometabolic traits and may discover new genes which have been missed by single variant genetic association analyses.
Our current proposal focuses on TWAS analyses, where genetic variants are analyzed in aggregate to derive a measure of predicted gene expression. We would like to extend this project to also include selected GWAS analyses, where variants are analyzed one by one, for the same traits. This will be used both for novel variant discovery with hematological traits (using a new set of reference haplotypes from the National Heart, Lung, and Blood Institute Trans-Omics for Precision Medicine (TOPMed) initiative (>100,000 haplotypes) to improve imputation of low frequency variants in African ancestry UK Biobank participants), likely in conjunction with TWAS, and for smaller scale replication efforts for ongoing genetic analyses in other consortia (such as the Population Architecture using Genomics and Epidemiology (PAGE) group). This will not require any additional data not requested in our current application, and will not pose any additional risks to research participants. This will not require any additional data not requested in our current application, except for already requested cardiometabolic traits such as lipid and glycemic measures which are not yet available, and will not pose any additional risks to research participants.
In addition to our current proposal focused on TWAS and selected GWAS analyses, we propose to further extend this project to include selected phenome!wide association studies (PheWAS). Described as a “reverse genome-wide association study”, a PheWAS elucidates links between suspected disease-causing genetic variants and a broad range of other phenotypes, which will allow us to further strengthen our results by replicating and validating selected GWAS and TWAS identified variants, while discovering novel potential genotype-phenotype associations. In this capacity, a PheWAS will allow us to fully exploit UK Biobank’s rich phenotypic data that are ascertained from multiple sources including linkage to various health-related records as well as the self-assessment questionnaire and measured biomarkers. Our proposed initial working example would involve examining genetic variants alone and in aggregate that affect lipoprotein(a) [Lp(a)], a highly atherogenic lipoprotein. In addition to its widely reported effects on CVD (e.g. increasing myocardial infarction [MI], heart failure, and venous thromboembolism risk), Lp(a) also has been associated with type 2 diabetes, liver disease, renal disease and breast cancer. Hypothesizing that Lp(a) harbors other yet-detected effects, we propose a large-scale PheWAS that spans phenotype domains including cardiovascular, cardiometabolic, cancer, kidney, psychiatric, ocular, and dermatological diseases and phenotypes by leveraging a genetically inferred Lp(a) to ensure temporality in exposure effects and limit bias from potential confounding. Based on the currently available GWAS literature for studies of Lp(a), we plan to evaluate the associations between the variants affecting the Lp(a) concentration and a broad spectrum of diseases either individually or in aggregate using a genetic risk score approach to increase the predictive power for association testing. By considering the “phenome” by which Lp(a) may exert its effects, we anticipate generating novel insights into biologic mechanisms by which this independent risk factor for atherosclerosis-related events influences health and disease. PheWAS for Lp(a) associated variants, as well as other sets of variants, can illuminate our understanding of molecular targets of relevant drugs as well as their potential unintended effects. This will require additional data on linked health outcomes and self-reported medical conditions besides the quantitative traits, such as lipid and glycemic measures, and ICD9/10 codes that were previously requested.
Last updated Jun 10, 2019